Influence of vocalic context on perception of the [zh]-[s] distinction.

Three experiments investigated the conditions under which fricative perception is influenced by following vocalic context. In Experiment 1, a reaction-time task, listeners showed no such influences , suggesting that they reached decisions about the fricative category before processing the vocalic context. In Experiment 2, a fixed-standard AX discrimination task employing synthetic fricative noises from a [5]-[s] continuum, listeners successfully discriminated fricative noises in isolation but shifted to a phonetic (categorical) mode of perception when vocalic context was added. Their response patterns changed systematically wi th the nature. of the context. In Experiment 3, the subjects listened first to pairs of isolated noises immediately followed by the same noises in context. When, subsequently, only noises in context were presented for discrimination, most of the subjects performed noncategorically and were no longer influenced by different vocalic contexts. These experiments demonstrate the availability of different perceptual strategies in listening to speech. In a recent study (Repp, 1980a), I used synthetic noises from a [5 ]-[s] continuum, followed by vocalic portions known to influence the location of the [)]-[s] boundary in an identification test. The stimuli were presented in AXB and fixed-standard AX discrimination tasks. The majority of naive subjects perceived these fricative-vowel syllables fairly categorically in both tasks; that is, discrimination functions followed the patterns predicted from identification scores and showed shifts contingent on the nature of the vocalic portion. However, two subjects achieved much better discrimination scores than the rest, and so did three experienced listeners who participated in the AX task. These listeners, who (judging from their higher accuracy, pattern of responses, and subjective reports) successfully followed the nonphonetic strategy of restricting attention to the spectral properties of the fricative noise, were not influenced by different vocalic contex ts. These resul ts supported the hypothesis that influences of vocalic context on fricative identification are tied to a phonetic mode of perception. EXPERIMENT 1 The experiment just summarized suggests that, when listening to fricativ~ vowel syllables in a phonetic mode, subjects process the vocalic portion Acknowledgment:.